A novel GIS-based ensemble technique for flood susceptibility mapping using evidential belief function and support vector machine: Brisbane, Australia

PeerJ

article

Mahyat Shafapour Tehrany¹  Lalit Kumar¹  Farzin Shabani¹ 

[1] School of Environmental and Rural Science, University of New England;Geospatial Science, School of Science, RMIT University;ARC Centre of Excellence for Australian Biodiversity and Heritage, Global Ecology, College of Science and Engineering, Flinders University of South Australia;Department of Biological Sciences, Macquarie University

关键词: Flood susceptibility mapping;    Support vector machine;    Evidential belief function;    Ensemble modeling;   

DOI  :  10.7717/peerj.7653

学科分类：社会科学、人文和艺术（综合）

来源: Inra

【 摘 要 】

In this study, we propose and test a novel ensemble method for improving the accuracy of each method in flood susceptibility mapping using evidential belief function (EBF) and support vector machine (SVM). The outcome of the proposed method was compared with the results of each method. The proposed method was implemented four times using different SVM kernels. Hence, the efficiency of each SVM kernel was also assessed. First, a bivariate statistical analysis using EBF was performed to assess the correlations among the classes of each flood conditioning factor with flooding. Subsequently, the outcome of the first stage was used in a multivariate statistical analysis performed by SVM. A highest prediction accuracy of 92.11% was achieved by an ensemble EBF-SVM—radial basis function method; the achieved accuracy was 7% and 3% higher than that offered by the individual EBF method and the individual SVM method, respectively. Among all the applied methods, both the individual EBF and SVM methods achieved the lowest accuracies. The reason for the improved accuracy offered by the ensemble methods is that by integrating the methods, a more detailed assessment of the flooding and conditioning factors can be performed, thereby increasing the accuracy of the final map.

【 授权许可】

CC BY